text stringlengths 300 320k | source stringlengths 52 154 |
|---|---|
# New report: A review of the empirical evidence for existential risk from AI via misaligned power-seeking
Visiting researcher Rose Hadshar recently published [a review of some evidence for existential risk from AI, focused on empirical evidence for misalignment and power seeking](https://arxiv.org/pdf/2310.18244.pdf)... | https://www.lesswrong.com/posts/3HfpCmKX7LJH5eTxQ/new-report-a-review-of-the-empirical-evidence-for |
# On Complexity Science
I have a long and confused love-hate relationship with the field of complex systems. People there never want to give me a simple, straightforward explanation about what its about, and much of what they say sounds a lot like woo ("edge of chaos" anyone?). But it also seems to promise a lot! This... | https://www.lesswrong.com/posts/PtMtMBHRZgHuup8sS/on-complexity-science |
# Partial value takeover without world takeover
People around me are very interested in AI taking over the world, so a big question is under what circumstances a system might be able to do that—what kind of capabilities could elevate an entity above the melange of inter-agent conflict and into solipsistic hegemony?
W... | https://www.lesswrong.com/posts/tJpwjpWtxYFENdsA3/partial-value-takeover-without-world-takeover |
# End-to-end hacking with language models
*Cross-posted from* [*https://tchauvin.com/end-to-end-hacking-with-language-models*](https://tchauvin.com/end-to-end-hacking-with-language-models)
*Produced as part of the SERI ML Alignment Theory Scholars Program - Summer 2023 Cohort.*
*Thanks to JS Denain and Léo Grinsztaj... | https://www.lesswrong.com/posts/jqXZzwvDWJZ3yAvYY/end-to-end-hacking-with-language-models |
# On the 2nd CWT with Jonathan Haidt
It was clear within the first ten minutes this would be a rich thread to draw from. In my childhood and education roundups, and of course with my own kids, I have been dealing with the issues Haidt talks about in his new book, [The Anxious Generation](https://www.penguinrandomhouse... | https://www.lesswrong.com/posts/6hciEN9DGsS8CEuox/on-the-2nd-cwt-with-jonathan-haidt |
# Koan: divining alien datastructures from RAM activations
*[Metadata: crossposted from [https://tsvibt.blogspot.com/2024/04/koan-divining-alien-datastructures-from.html](https://tsvibt.blogspot.com/2024/04/koan-divining-alien-datastructures-from.html).]*
Exploring the ruins of an alien civilization, you find what ap... | https://www.lesswrong.com/posts/unCG3rhyMJpGJpoLd/koan-divining-alien-datastructures-from-ram-activations |
# Exploring Whole Brain Emulation
I've been dedicating a fair amount of my time recently to investigating
whole brain emulation (WBE).
As computational power continues to grow, the feasibility of emulating a
human brain at a reasonable speed becomes increasingly plausible.
While the connectome data alone seems... | https://www.lesswrong.com/posts/4MeomxkQ8KEzsLKTW/exploring-whole-brain-emulation |
# How does the ever-increasing use of AI in the military for the direct purpose of murdering people affect your p(doom)?
I haven't personally heard a lot of recent discussions about it, which is strange considering that both startups like Andruil and Palantir are developing systems for military use, OpenAI recently de... | https://www.lesswrong.com/posts/ebgazvWG5Kxuy4Wff/how-does-the-ever-increasing-use-of-ai-in-the-military-for |
# Privacy and writing
**Epistemic status**: N=1
I've always written several thousand words a day in a private Google doc about anything that came to mind. Only recently have I started publishing to LessWrong. It's a long and arduous process for me, too slow to be worth the effort usually. [^wqf0jyw58ri] Still, publi... | https://www.lesswrong.com/posts/eA5eAexedL7MhPgCZ/privacy-and-writing |
# Measuring Predictability of Persona Evaluations
*This work was done by Thee Ho as part of the Athena 1.0 mentorship program under Evan Hubinger. Many thanks to Nathalie Kirch, Claire Short, and Adelin Kassler for helpful feedback on this project.*
Overview
=========
We are interested in understanding the difficul... | https://www.lesswrong.com/posts/KCPYzWvo8z5nJWajv/measuring-predictability-of-persona-evaluations |
# My intellectual journey to (dis)solve the hard problem of consciousness
**Epistemological status:** At least a fun journey... | https://www.lesswrong.com/posts/Ru2cDrre6D4gkf734/my-intellectual-journey-to-dis-solve-the-hard-problem-of |
# The 2nd Demographic Transition
Birth rates in the developed world are below replacement levels and [global fertility](https://ourworldindata.org/fertility-rate) is not far behind. Sub-replacement fertility leads to exponentially decreasing population. Our [best models of economic growth](https://www.maximum-progress... | https://www.lesswrong.com/posts/ncbmN2qmAacwWtjpE/the-2nd-demographic-transition |
# "Fractal Strategy" workshop report
I just ran a workshop teaching the [rationality](https://www.lesswrong.com/posts/pZrvkZzL2JnbRgEBC/feedbackloop-first-rationality) [concepts](https://www.lesswrong.com/posts/FEMWS79AFTmKq2iJK/rationality-research-report-towards-10x-ooda-looping) I've developed this year.
If you'r... | https://www.lesswrong.com/posts/h5mDx2Mt2P5m9v588/fractal-strategy-workshop-report |
# Conflict in Posthuman Literature
Grant Snider [created this comic](http://www.incidentalcomics.com/2014/05/conflict-in-literature.html) (which became a meme):
. Why is that?
---
6-month US Treasuries are paying around 5.3% interest. Japanese government bonds are paying about 0%. That being the case, you can borrow yen, trade it... | https://www.lesswrong.com/posts/yF3nnfYdAoHPAzNkH/on-the-dollar-yen-exchange-rate |
# The Poker Theory of Poker Night
_Link to my own article. I removed the explanation of EV since I assume on LW that's not necessary._
A group of friends and I occasionally like to get together to play Poker. Yet
something keeps happening that I have observed time and again with these kinds
of group gatherings: It is... | https://www.lesswrong.com/posts/Lax72t8c44kxFh4iB/the-poker-theory-of-poker-night |
# Applications Open: Elevate Your Mental Wellbeing with Rethink Wellbeing's CBT Program
Do you want to:
* become more fullfilled, resilient, and productive?
* practice evidence-based tools for self-management, to deal with blockers and stressors such as low concentration, motivation, mood, and self-esteem?
* em... | https://www.lesswrong.com/posts/Lw2k5d3ACEfNAnwC5/applications-open-elevate-your-mental-resilience-with |
# A Dozen Ways to Get More Dakka
As [the dictum goes](https://www.lesswrong.com/posts/z8usYeKX7dtTWsEnk/more-dakka), “If it helps but doesn’t solve your problem, perhaps you’re not using enough.” But I still find that I’m sometimes not using enough effort, not doing enough of what works, simply put, not using enough d... | https://www.lesswrong.com/posts/KgFNwBuaDpfGSJktM/a-dozen-ways-to-get-more-dakka |
# Normalizing Sparse Autoencoders
TL;DR
=====
Sparse autoencoders (SAEs) presents us a promising direction towards automating mechanistic interpretability, but it not without flaws. One known issue of the original sparse autoencoders is the [*feature suppression*](https://www.lesswrong.com/posts/3JuSjTZyMzaSeTxKk/add... | https://www.lesswrong.com/posts/3ZCKSArYwgg9P4hqQ/normalizing-sparse-autoencoders |
# Math-to-English Cheat Sheet
Say you've learnt math in your native language which is not English. Since then you've also read math in English and you appreciate the near universality of mathematical notation. Then one day you want to discuss a formula in real life and you realize you don't know how to pronunce "$a_n$... | https://www.lesswrong.com/posts/yEsuwCugokgpAQyYD/math-to-english-cheat-sheet |
# Gated Attention Blocks: Preliminary Progress toward Removing Attention Head Superposition
*This work represents progress on removing attention head superposition. We are excited by this approach but acknowledge there are currently various limitations. In the short term, we will be working on adjacent problems are ex... | https://www.lesswrong.com/posts/kzc3qNMsP2xJcxhGn/gated-attention-blocks-preliminary-progress-toward-removing-1 |
# Measuring Learned Optimization in Small Transformer Models
*This is original, independent research carried out in March and April of 2024.*
*, I introduced a framework for applying scenario planning to AI x-risk. My colleagues at Convergence followed up with posts* [*defining*](https://forum.e... | https://www.lesswrong.com/posts/2aarzb6vwJg9f9yQR/investigating-the-role-of-agency-in-ai-x-risk |
# CEA seeks co-founder for AI safety group support spin-off
Summary
-------
* CEA is currently[ inviting expressions of interest](https://forms.gle/ZVL1enuG4sUBYJQe8) for co-founding a promising new project focused on providing non-monetary support to AI safety groups. We’re also [receiving recommendations for the ... | https://www.lesswrong.com/posts/KFybgPDTaerANipEb/cea-seeks-co-founder-for-ai-safety-group-support-spin-off |
# How We Picture Bayesian Agents
I think that when most people picture a Bayesian agent, they imagine a system which:
* Enumerates every possible state/trajectory of “the world”, and assigns a probability to each.
* When new observations come in, loops over every state/trajectory, checks the probability of the ob... | https://www.lesswrong.com/posts/TiBsZ9beNqDHEvXt4/how-we-picture-bayesian-agents |
# Pandemic Identification Simulator
At my [day job](https://naobservatory.org/) I work on identifying potential pandemics sooner, so we have more time to respond. I recently made a simulator which pulls a [lot of](https://www.jefftk.com/p/weekly-incidence-including-delay) [things I've](https://www.jefftk.com/p/wastewa... | https://www.lesswrong.com/posts/KvSgty2jY7XrpJk5r/pandemic-identification-simulator |
# Fermenting Form
Application forms are hard to write. Questions like: “Tell us about yourself” or “What are your strengths and weaknesses?” are tiring to address usefully in 200 words. Getting stuck on... | https://www.lesswrong.com/posts/FHi2fq9mpBaNuDirj/fermenting-form |
# [Closed] PIBBSS is hiring in a variety of roles (alignment research and incubation program)
PIBBSS is looking to expand its team and is running work trials for new team members (primarily) in April, May and early June. If you’re interested in joining a nimble team focused on AI safety research, field-building and in... | https://www.lesswrong.com/posts/e6YukTAMzB7krHxRW/closed-pibbss-is-hiring-in-a-variety-of-roles-alignment |
# Medical Roundup #2
Previously: [#1](https://thezvi.substack.com/p/medical-roundup-1)
It feels so long ago that Covid and health were my beat, and what everyone often thought about all day, rather than AI. Yet the beat goes on. With Scott Alexander at long last giving us what I expect to be effectively the semi-fina... | https://www.lesswrong.com/posts/wfz47Ez2r4rQZuYBY/medical-roundup-2 |
# D&D.Sci: The Mad Tyrant's Pet Turtles [Evaluation and Ruleset]
This is a followup to [the D&D.Sci post](https://www.lesswrong.com/posts/t9dGiqj9M3Ataaueo/d-and-d-sci-the-mad-tyrant-s-pet-turtles) I made ten days ago; if you haven’t already read it, you should do so now before spoiling yourself.
[Here](https://h-b-p... | https://www.lesswrong.com/posts/LDFAjLXDSWRpSgxj5/d-and-d-sci-the-mad-tyrant-s-pet-turtles-evaluation-and |
# Apply to LASR Labs: a London-based technical AI safety research programme
Edit: **Applications for this round are now closed!** If you are interested in future rounds, you can [express interest here](https://airtable.com/appbzbkQ3OwRBaojt/shruJmwbbk07e1i7y).
TLDR; [apply](https://airtable.com/appbzbkQ3OwRBaojt/shr... | https://www.lesswrong.com/posts/Sqjoxk74wvrcdpBxr/apply-to-lasr-labs-a-london-based-technical-ai-safety |
# Ophiology (or, how the Mamba architecture works)
The following post was made as part of Danielle's MATS work on doing circuit-based mech interp on Mamba, mentored by Adrià Garriga-Alonso. It's the first in a sequence of posts about finding an IOI circuit in Mamba/applying [ACDC](https://arxiv.org/abs/2304.14997) to ... | https://www.lesswrong.com/posts/TYLQ8gAMAmpeFcwXN/ophiology-or-how-the-mamba-architecture-works |
# How to accelerate recovery from sleep debt with biohacking?
I have at least 40 hours of sleep debt from a polyphasic sleep schedule and attending hackathons. This number is a conservative estimate. Has anyone here researched the neurobiology of sleep deprivation? What can I do to recover quickly? | https://www.lesswrong.com/posts/bQDoyQNKAaFJnT78p/how-to-accelerate-recovery-from-sleep-debt-with-biohacking |
# How I select alignment research projects
Youtube Video
-------------
Recently, I was interviewed by Henry Sleight and Mikita Balesni about how I select alignment research projects. Below is the slightly cleaned up transcript for the YouTube video.
Introductions
-------------
**Henry Sleight**: How about you two i... | https://www.lesswrong.com/posts/7GmDs4BqrFW3kk4nP/how-i-select-alignment-research-projects |
# Apply to the Pivotal Research Fellowship (AI Safety & Biosecurity)
The Swiss Existential Risk Initiative (CHERI) is now called *Pivotal Research,* and the CHERI research fellowship is now the *Pivotal Research Fellowship*.
| [Apply](https://airtable.com/app79ObszqhBUadQZ/pagH7yxKfZnT9EW2z/form) for the [Pivotal Res... | https://www.lesswrong.com/posts/5YNFJLQgANn2QqotY/apply-to-the-pivotal-research-fellowship-ai-safety-and |
# (Rational) Decision-Making In Wartime
The martial art of rationality provides for seeing what the science of rationality means for ourselves, for our daily inner life. If so, I wonder whether war complicates exercising the muscles of sound reasoning or alters what rationality means for me. Along with the systemati... | https://www.lesswrong.com/posts/ssDyKqzNaxpazG9ki/rational-decision-making-in-wartime |
# RTFB: On the New Proposed CAIP AI Bill
A New Bill Offer Has Arrived
[Center for AI Policy proposes](https://twitter.com/aipolicyus/status/1777690683190509602) [a concrete actual model bill for us to look at.](https://assets.caip.org/caip/RAAIA%20%28March%202024%29.pdf)
Here was their announcement:
> WASHINGTON – ... | https://www.lesswrong.com/posts/SQ9wDmsELBmA4Lega/rtfb-on-the-new-proposed-caip-ai-bill |
# Is LLM Translation Without Rosetta Stone possible?
Suppose astronomers detect a binary radio signal, an alien message, from a star system many light years away. The message contains a large text dump (conveniently, about GPT-4 training text data sized) composed in an alien language. Let's call it Alienese.[^1]
Unfo... | https://www.lesswrong.com/posts/J3zA3T9RTLkKYNgjw/is-llm-translation-without-rosetta-stone-possible |
# Reverse Regulatory Capture
Reverse regulatory capture is an advanced move where industry interests cry "regulatory capture" in order to oppose regulation. Generally, the more skeptical people are about corporate power and the vulnerability of regulators to influence, the more worried they are about regulatory captur... | https://www.lesswrong.com/posts/xpyvJ76brChicdfrC/reverse-regulatory-capture |
# Work ethic after 2020?
It seems to me that my work ethic / tendency to akrasia / etc. has changed a lot since 2020, and many people around me seem to agree that it’s also become harder for them to keep working on things they don’t really enjoy doing.
But I find myself very confused about this topic, so I’d like to ... | https://www.lesswrong.com/posts/XYfFG84gDoEuqRww3/work-ethic-after-2020 |
# AI #59: Model Updates
Claude uses tools now. Gemini 1.5 is available to everyone and Google promises more integrations. GPT-4-Turbo gets substantial upgrades. Oh and new model from Mistral, TimeGPT for time series, and also new promising song generator. No, none of that adds up to GPT-5, but everyone try to be a lit... | https://www.lesswrong.com/posts/hQaxcitYgKjJqMdps/ai-59-model-updates |
# Announcing Atlas Computing
[Atlas Computing](https://atlascomputing.org/) is a new nonprofit working to collaboratively advance AI capabilities that are asymmetrically risk-reducing. Our work consists of building scoped prototypes and creating an ecosystem around [@davidad](https://www.lesswrong.com/users/davidad?me... | https://www.lesswrong.com/posts/dKD47KMqvYAB7Zkmv/announcing-atlas-computing |
# A Gentle Introduction to Risk Frameworks Beyond Forecasting
*This was originally posted on Nathaniel's and Nuno's substacks (*[*Pending Survival*](https://nathanielcooke.substack.com/p/a-gentle-introduction-to-risk-frameworks) *and* [*Forecasting Newsletter*](https://forecasting.substack.com/p/a-gentle-introduction-... | https://www.lesswrong.com/posts/ae3HecTe2uKscabPe/a-gentle-introduction-to-risk-frameworks-beyond-forecasting |
# Experience Report - ML4Good AI Safety Bootcamp
In this post I reflect on my experience in participating at a [ML4Good](https://www.ml4good.org/) (UK, March 2024) bootcamp. I am writing this mainly for my own benefit - to reflect and to start making a habit of engaging with the community on lesswrong. But also to hel... | https://www.lesswrong.com/posts/bqYXX9K28DGB7Azog/experience-report-ml4good-ai-safety-bootcamp |
# Structured Transparency: a framework for addressing
use/mis-use trade-offs when sharing information
Andrew Trask together with Ben Garfinkel, Allan Dafoe and Eric Drexler (and others I know less well) released a paper on transparency and infohazards that seemed interesting on a first skim:
> **Abstract:** Successf... | https://www.lesswrong.com/posts/35djfj67zu9E9TDRH/structured-transparency-a-framework-for-addressing-use-mis |
# Ackshually, many worlds is wrong
*Thank you to Justis Mills for proofreading and feedback. This post can also be found* [*on my substack*](https://tailcalled.substack.com/p/ackshually-many-worlds-is-wrong)*.*
I mentioned that [I disagree with the many worlds interpretation of quantum mechanics in a comment](https:/... | https://www.lesswrong.com/posts/Y6LhXdGfwsAStMuhr/ackshually-many-worlds-is-wrong |
# AXRP Episode 27 - AI Control with Buck Shlegeris and Ryan Greenblatt
[YouTube link](https://youtu.be/gQbCO6zGRiI)
A lot of work to prevent AI existential risk takes the form of ensuring that AIs don’t want to cause harm or take over the world—or in other words, ensuring that they’re aligned. In this episode, I talk... | https://www.lesswrong.com/posts/7vRiJozcE2AjcRm7J/axrp-episode-27-ai-control-with-buck-shlegeris-and-ryan |
# Leave No Context Behind - A Comment
[Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention](https://arxiv.org/abs/2404.07143) by Tsendsuren Munkhdalai, Manaal Faruqui, and Siddharth Gopal of Google.
This is a pre-print of a new LLM extension with what I'd call short-term memory, but... | https://www.lesswrong.com/posts/gcv86vXgQhbBDNGNx/leave-no-context-behind-a-comment |
# A D&D.Sci Dodecalogue
*Below is some advice on making* [*D&D.Sci scenarios*](https://www.lesswrong.com/tag/d-and-d-sci)*. I’m mostly yelling it in my own ear, and you shouldn’t take any of it as gospel; but if you want some guidance on how to run your first game, you may find it helpful.*
**1\. The scoring function... | https://www.lesswrong.com/posts/c5xAbkQoanAueBTkx/a-d-and-d-sci-dodecalogue |
# Interference Issues
I've been working on building an electronic harp mandolin ( [previously](https://www.jefftk.com/p/pluck-sensor-circuit), [previously](https://www.jefftk.com/p/mandolin-harp-sensor-placement), [previously](https://www.jefftk.com/p/prototyping-pluck-sensors)). I got it all hooked up, but unfortunat... | https://www.lesswrong.com/posts/Kem384QRdjuNpDot9/interference-issues |
# Report: Evaluating an AI Chip Registration Policy
*As part of our *[***Governance Recommendations Research Program***](https://www.convergenceanalysis.org/programs/governance-research)*, *[***Convergence Analysis***](https://www.convergenceanalysis.org/) *is publishing a series of deep-dive analyses on specific, upc... | https://www.lesswrong.com/posts/8u8x2bSpG9LLa8jfN/report-evaluating-an-ai-chip-registration-policy |
# UDT1.01: Plannable and Unplanned Observations (3/10)
**The Omnipresence of Unplanned Observations**
Time to introduce some more concepts. If an observation is "any data you can receive which affects your actions", then there seem to be two sorts of observations. A plannable observation is the sort of observation wh... | https://www.lesswrong.com/posts/kjMLK83vqpuLugst4/udt1-01-plannable-and-unplanned-observations-3-10 |
# AISN #33: Reassessing AI and Biorisk
Plus, Consolidation in the Corporate AI Landscape, and National Investments in AI
Welcome to the AI Safety Newsletter by the [Center for AI Safety](https://www.safe.ai/). We discuss developments in AI and AI safety. No technical background required.
Subscribe [here](https://news... | https://www.lesswrong.com/posts/8pBkrL9coqchqKi3r/aisn-33-reassessing-ai-and-biorisk-plus-consolidation-in-the |
# Generalized Stat Mech: The Boltzmann Approach
Context
=======
There's a common intuition that the tools and frames of statistical mechanics ought to generalize far beyond physics and, of particular interest to us, it feels like they ought to say a lot about agency and intelligence. But, in practice, attempts to app... | https://www.lesswrong.com/posts/keGhMbnLNdTAHa4bv/generalized-stat-mech-the-boltzmann-approach |
# Poker, Beef Wellington, and Mount Stupid
Introduction
------------
This post will hopefully serve to illustrate a common pattern I see, one of those things that you see everywhere once you figure it out. I would be surprised if most people who read this *haven't* had the same thoughts as me, but maybe the direction... | https://www.lesswrong.com/posts/zrtu2Qg4JNjkfwCmv/poker-beef-wellington-and-mount-stupid |
# MIRI's April 2024 Newsletter
The MIRI Newsletter is back in action after a hiatus since July 2022. To recap some of the biggest MIRI developments since then:
* MIRI released its [2024 Mission and Strategy Update](https://intelligence.org/2024/01/04/miri-2024-mission-and-strategy-update/), announcing a major shift... | https://www.lesswrong.com/posts/n22oXbeKDyt4ysDv8/miri-s-april-2024-newsletter |
# Things Solenoid Narrates
I spend *a lot* of time narrating various bits of EA/longtermist writing.
The resulting audio exists in many different places. Surprisingly often, people who really like one thing don't know about the other things. This seems bad.[^shn4hjix8r]
A few people have requested a feed to aggregat... | https://www.lesswrong.com/posts/7oQxHQeXsQZEcSAzQ/things-solenoid-narrates |
# Prompts for Big-Picture Planning
During my [metastrategy workshop](https://www.lesswrong.com/posts/h5mDx2Mt2P5m9v588/fractal-strategy-workshop-report), Day Two was focused on taking a step back and asking "okay, wait, what am I actually doing and why?". Choosing what area to focus, and what your mid-level strategy i... | https://www.lesswrong.com/posts/KLsKaywDDLRSLgdFC/prompts-for-big-picture-planning |
# Carl Sagan, nuking the moon, and not nuking the moon
In 1957, Nobel laureate microbiologist Joshua Lederberg and biostatician J. B. S. Haldane sat down together imagined what would... | https://www.lesswrong.com/posts/hwtt9zM3MxoKnwgbd/carl-sagan-nuking-the-moon-and-not-nuking-the-moon |
# Consequentialism is a compass, not a judge
**Tl;dr:** Consequentialism works as a compass for your actions, not as a judge of moral character.
The compass and the judge
-------------------------
A woman steps onto a crowded bus, trips on a sitting man's outstretched foot, and breaks her arm. The Everett branches s... | https://www.lesswrong.com/posts/JumyfYQaJkWnjCcEr/consequentialism-is-a-compass-not-a-judge |
# My experience at ML4Good AI Safety Bootcamp
TLDR
====
* Experience was fantastic: I learnt a lot, made new friends and potential collaborators, and it was the step-change I needed to kickstart my contributions to AI safety.
* Big lesson for me is that mechanistic interpretability (mech interp) is not the only w... | https://www.lesswrong.com/posts/ggQ4hFaL9EimsZ8ud/my-experience-at-ml4good-ai-safety-bootcamp |
# What convincing warning shot could help prevent extinction from AI?
> *\- Tell me father, when is the line*
>
> *where en... | https://www.lesswrong.com/posts/RYx6cLwzoajqjyB6b/what-convincing-warning-shot-could-help-prevent-extinction |
# Text Posts from the Kids Group: 2020
Another round of liberating kid posts from [Facebook](https://www.jefftk.com/p/making-groups-for-kid-pictures). For reference, in 2020 Lily turned 6 and Anna turned 4.
(Some of these were from me; some were from Julia. Ones saying "me" could mean either of us.)
[2020-01-01](htt... | https://www.lesswrong.com/posts/3aonzw5HZqpDfBZxC/text-posts-from-the-kids-group-2020 |
# [Cosmology Talks] New Probability Axioms Could Fix Cosmology's Multiverse (Partially) - Sylvia Wenmackers
> Sylvia is a philosopher of science. Her focus is probability and she has worked on a few theories that aim to extend and modify the standard axioms of probability in order to tackle paradoxes related to infini... | https://www.lesswrong.com/posts/eoyNS4nTeYoArBgJv/cosmology-talks-new-probability-axioms-could-fix-cosmology-s |
# UDT1.01 Essential Miscellanea (4/10)
This is the post with some needed concepts and discussion that didn't cleanly fit into any other section, so it might be a bit of a rambly mess.
Specifically, this post splits into two parts. One is assorted musings about when to defer to past-you vs current-you when making dec... | https://www.lesswrong.com/posts/jsNGHqaMwQN76pxNP/udt1-01-essential-miscellanea-4-10 |
# A quick experiment on LMs’ inductive biases in performing search
TL;DR: Based on a toy setting, GPT-3.5-turbo and GPT-4-turbo are best at search (by which I mean computing an argmax) when using chain-of-thought, but neither of them can do *internal *search when forced to work from memory over only a few token positi... | https://www.lesswrong.com/posts/XaKLjyDejtXDoRAzL/a-quick-experiment-on-lms-inductive-biases-in-performing |
# A High Decoupling Failure
High-decoupling vs low-decoupling or decoupling vs contextualizing refers to two different cultural norms, cognitive skills, or personal dispositions that change the way people approach ideas.
> [High-decouplers](https://everythingstudies.com/2018/04/26/a-deep-dive-into-the-harris-klein-co... | https://www.lesswrong.com/posts/hXS2iZ8H3Xep6JPxv/a-high-decoupling-failure |
# Clipboard Filtering
Here's a pattern I find pretty useful:
pbpaste | some_command | pbcopy
For example:
* Converting spaces to tabs, for pasting into a spreadsheet program: `pbpaste | tr ' ' '\t' | pbcopy`
* Converting tabs and newlines to html table formatting: `pbpaste | sed 's/^/<tr><td>/' | sed 's/\t... | https://www.lesswrong.com/posts/THrvd53ktGirsRBng/clipboard-filtering |
# Anthropic AI made the right call
I've seen a number of people criticize Anthropic for releasing Claude 3 Opus, with arguments along the lines of:
> Anthropic said they weren't going to push the frontier, but this release is clearly better than GPT-4 in some ways! They're betraying their mission statement!
I think t... | https://www.lesswrong.com/posts/BaLAgoEvsczbSzmng/anthropic-ai-made-the-right-call |
# Reconsider the anti-cavity bacteria if you are Asian
**Update: further research has led me to believe that people of all races should test themselves for ALDH deficiency before using Lumina. Even if you don't exhibit AFR symptoms when drinking alcohol, your ALDH activity may still be decreased.**
Many people in the... | https://www.lesswrong.com/posts/jGu4nLgQYwfsoxddu/reconsider-the-anti-cavity-bacteria-if-you-are-asian |
# Monthly Roundup #17: April 2024
As always, a lot to get to. This is everything that wasn’t in any of the other categories.
#### Bad News
[You might have to find a way to actually enjoy the wor](https://twitter.com/gdb/status/1771058046573392177)k.
> Greg Brockman (President of OpenAI): Sustained great work often ... | https://www.lesswrong.com/posts/cbkJWkKWvETwJqoj2/monthly-roundup-17-april-2024 |
# Taking into account preferences of past selves
Problem
=======
“You’ll have a great time wherever you go to college!” I constantly hear this. From my parents, my friends’ parents, my guidance counselor, and my teachers. I don’t doubt it. I’m sure I’ll have a lot of fun wherever I go. Since I’m trying to be very int... | https://www.lesswrong.com/posts/jW8PKh29xZ8Atpsrt/taking-into-account-preferences-of-past-selves |
# Effectively Handling Disagreements - Introducing a New Workshop
On May 25th, 2023, someone posted a [review of How Minds Change](https://www.lesswrong.com/posts/D2GrrrrfipHWPJSHh/book-review-how-minds-change) on LessWrong. It talked about Street Epistemology, Deep Canvassing, and Smart Politics, ways of handling dis... | https://www.lesswrong.com/posts/peeibJXeqPQ4zXkvj/effectively-handling-disagreements-introducing-a-new |
# Experiments with an alternative method to promote sparsity in sparse autoencoders
### **Summary**
I experimented with alternatives to the standard L1 penalty used to promote sparsity in sparse autoencoders (SAEs). I found that including terms based on an alternative differentiable approximation of the feature spars... | https://www.lesswrong.com/posts/cYA3ePxy8JQ8ajo8B/experiments-with-an-alternative-method-to-promote-sparsity |
# An evaluation of circuit evaluation metrics
*This work was produced as part of the*[* ML Alignment & Theory Scholars Program*](https://www.matsprogram.org/) *\- Winter 2023-24 Cohort, with mentorship from Adrià Garriga-Alonso. Equal contribution by Niels and Iván.*
Summary
-------
In this post we analyse the circu... | https://www.lesswrong.com/posts/4YiFiPGYMazidFgQ2/an-evaluation-of-circuit-evaluation-metrics-1 |
# My experience using financial commitments to overcome akrasia
About a year ago I decided to try using [one](https://www.beeminder.com/) [of](https://bossasaservice.com/) [those](https://taskratchet.com/) [apps](https://complice.co/) where you tie your goals to some kind of financial penalty. The specific one I tried... | https://www.lesswrong.com/posts/DRrAMiekmqwDjnzS5/my-experience-using-financial-commitments-to-overcome |
# The argument for near-term human disempowerment through AI
**Author**: Leonard Dung
**Abstract**: Many researchers and intellectuals warn about extreme risks from artificial intelligence. However, these warnings typically came without systematic arguments in support. This paper provides an argument that AI will... | https://www.lesswrong.com/posts/ALrJR2wacBBNBQnqD/the-argument-for-near-term-human-disempowerment-through-ai |
# Announcing SPAR Summer 2024!
**The **[**Supervised Program for Alignment Research**](https://supervisedprogramforalignment.org/) **(SPAR) is open to applications for summer 2024!** SPAR connects mentors and mentees for three-month AI safety and governance research projects. **The **[**Summer 2024 cycle of SPAR**](ht... | https://www.lesswrong.com/posts/J8wCSsu9dkMEEhRva/announcing-spar-summer-2024 |
# Essay competition on the Automation of Wisdom and Philosophy — $25k in prizes
With AI Impacts, we’re pleased to announce an essay competition on the automation of wisdom and philosophy. Submissions are due by July 14th, 2024. The first prize is $10,000, and there is a total of $25,000 in prizes available.
[Submit ... | https://www.lesswrong.com/posts/52ygLry5KCdvxY6zn/essay-competition-on-the-automation-of-wisdom-and-philosophy |
# {Book Summary} The Art of Gathering
This is a Book Review & Summary of [The Art of Gathering: How We Meet and Why It Matters](https://www.goodreads.com/book/show/37424706-the-art-of-gathering?ac=1&from_search=true&qid=QGXBcFLw1I&rank=3) by Priya Parker.
**Rating**: 4/5
I've pulled the main insights and actionable... | https://www.lesswrong.com/posts/Cc8HXLSqf2TuvyeJv/book-summary-the-art-of-gathering |
# What should the EA community learn from the FTX / SBF disaster? An in-depth discussion with Will MacAskill on the Clearer Thinking podcast
In this new podcast episode, I discuss with Will MacAskill what the Effective Altruism community can learn from the FTX / SBF debacle, why Will has been limited in what he co... | https://www.lesswrong.com/posts/pSi9HcqhG2QkzvDRj/what-should-the-ea-community-learn-from-the-ftx-sbf-disaster |
# Creating unrestricted AI Agents with Command R+
TL;DR There currently are capable open-weight models which can be used to create simple unrestricted bad agents. They can perform tasks end-to-end such as [searching for negative information](https://www.lesswrong.com/posts/4vPZgvhmBkTikYikA/creating-bad-agents-with-co... | https://www.lesswrong.com/posts/4vPZgvhmBkTikYikA/creating-unrestricted-ai-agents-with-command-r |
# Paul Christiano named as US AI Safety Institute Head of AI Safety
> U.S. Secretary of Commerce Gina Raimondo announced today additional members of the executive leadership team of the U.S. AI Safety Institute (AISI), which is housed at the National Institute of Standards and Technology (NIST). Raimondo named **Paul ... | https://www.lesswrong.com/posts/63X9s3ENXeaDrbe5t/paul-christiano-named-as-us-ai-safety-institute-head-of-ai |
# Transformers Represent Belief State Geometry in their Residual Stream
*Produced while being an affiliate at PIBBSS*[^nla6r2hm9te]*. The work was done initially with funding from a Lightspeed Grant, and then continued while at PIBBSS. Work done in collaboration with* [*@Paul Riechers*](https://www.lesswrong.com/users... | https://www.lesswrong.com/posts/gTZ2SxesbHckJ3CkF/transformers-represent-belief-state-geometry-in-their |
# Spending Update 2024
I'm generally a pretty big fan of transparency, and one way I try to promote this is writing up our finances every few years ( [2022](https://www.jefftk.com/p/spending-update-2022), [2020](https://www.jefftk.com/p/spending-update-2020), [2018](https://www.jefftk.com/p/spending-update-2018), [201... | https://www.lesswrong.com/posts/APwfRXs8SMEx6KjvS/spending-update-2024 |
# Mid-conditional love
People talk about unconditional love and conditional love. Maybe I’m out of the loop regarding the great loves going on around me, but my guess is that love is extremely rarely unconditional. Or at least if it is, then it is either very broadly applied or somewhat confused or strange: if you lov... | https://www.lesswrong.com/posts/rwBbTaN9WLfCA7MAo/mid-conditional-love |
# When is a mind me?
xlr8harder writes:
> In general I don’t think an uploaded mind is you, but rather a copy. But one thought experiment makes me question this. A Ship of Theseus concept where individual neurons are replaced one at a time with a nanotechnological functional equivalent.
>
> Are you still you?
Presu... | https://www.lesswrong.com/posts/zPM5r3RjossttDrpw/when-is-a-mind-me |
# Claude 3 Opus can operate as a Turing machine
Posted on Twitter:
> * Opus can operate as a Turing machine.
> * given only existing tapes, it learns the rules and computes new sequences correctly.
> * 100% accurate over 500+ 24-step solutions (more tests running).
> * for 100% at 24 steps, the input tapes... | https://www.lesswrong.com/posts/wbjXQtiWMcfo4EpM4/claude-3-opus-can-operate-as-a-turing-machine |
# Childhood and Education Roundup #5
For this iteration I will exclude discussions involving college or college admissions.
There has been a lot of that since the last time I did one of these, along with much that I need to be careful with lest I go out of my intended scope. It makes sense to do that as its own treat... | https://www.lesswrong.com/posts/s34ingEzvajpFPaaD/childhood-and-education-roundup-5 |
# FHI (Future of Humanity Institute) has shut down (2005–2024)
> Over time FHI faced increasing administrative headwinds within the Faculty of Philosophy (the Institute’s organizational home). Starting in 2020, the Faculty imposed a freeze on fundraising and hiring. In late 2023, the Faculty of Philosophy decided th... | https://www.lesswrong.com/posts/tu3CH22nFLLKouMKw/fhi-future-of-humanity-institute-has-shut-down-2005-2024 |
# Staged release
"Staged release" is regularly mentioned as *a good thing for frontier AI labs to do*. But I've only ever seen one analysis of staged release,[^uj9a8bk9zt] and the term's meaning has changed, becoming vaguer since the GPT-2 era.
This post is kinda a reference post, kinda me sharing my understanding/ta... | https://www.lesswrong.com/posts/RdsqJnmP8XxutLbso/staged-release |
# Moving on from community living
After 7 years at Deep End (and 4 more years in other group houses before that), Janos and I have moved out to live near a school we like and some lovely parks. The life change is bittersweet - we will miss living with our friends, but also look forward to a logistically simpler life w... | https://www.lesswrong.com/posts/KjNv9pzNcbcNaR5F3/moving-on-from-community-living |
# LLM Evaluators Recognize and Favor Their Own Generations
Self-evaluation using LLMs is used in [reward](https://arxiv.org/abs/1811.07871) [modeling](https://arxiv.org/abs/2009.01325), [model](http://ar... | https://www.lesswrong.com/posts/CeCKzsKABGKt9aabb/llm-evaluators-recognize-and-favor-their-own-generations |
# AXRP Episode 28 - Suing Labs for AI Risk with Gabriel Weil
[YouTube link](https://youtu.be/hdQDZo7jGBY)
How should the law govern AI? Those concerned about existential risks often push either for bans or for regulations meant to ensure that AI is developed safely - but another approach is possible. In this episode,... | https://www.lesswrong.com/posts/JQEeNgezmPBz6n2je/axrp-episode-28-suing-labs-for-ai-risk-with-gabriel-weil |
# Why Would Belief-States Have A Fractal Structure, And Why Would That Matter For Interpretability? An Explainer
Yesterday Adam Shai put up a [cool post](https://www.lesswrong.com/posts/gTZ2SxesbHckJ3CkF/transformers-represent-belief-state-geometry-in-their) which… well, take a look at the visual:
![](https://lh7-us.... | https://www.lesswrong.com/posts/mBw7nc4ipdyeeEpWs/why-would-belief-states-have-a-fractal-structure-and-why |
# Express interest in an "FHI of the West"
*TLDR: I am investigating whether to found a spiritual successor to* [*FHI*](https://www.futureofhumanityinstitute.org/)*, housed under* [*Lightcone Infrastructure*](https://lightconeinfrastructure.com/)*, providing a rich cultural environment and financial support to researc... | https://www.lesswrong.com/posts/ydheLNeWzgbco2FTb/express-interest-in-an-fhi-of-the-west |
# The Mom Test: Summary and Thoughts
I just finished reading [The Mom Test](https://www.momtestbook.com/) for the second time. I took "raw" notes [here](https://docs.google.com/document/d/1N7hBXErrd8ymZULPYtyNnQlmeATBLgvqbjqIJqgVgvc/edit?usp=sharing). In this post I'll first write up a bullet-point summary and then ra... | https://www.lesswrong.com/posts/fuktKYSvzFMTfPpeT/the-mom-test-summary-and-thoughts |
# UDT1.01: Logical Inductors and Implicit Beliefs (5/10)
One of the primary conceptual challenges of UDT is that, if future-you is going to be deferring to past-you about what to do in various circumstances, and past-you hasn't exhaustively thought through every possible circumstance ahead of time, that causes a tensi... | https://www.lesswrong.com/posts/GiHRBRxFaKKgDkr5p/udt1-01-logical-inductors-and-implicit-beliefs-5-10 |
# AI #60: Oh the Humanity
Many things this week did not go as planned.
Humane AI premiered its AI pin. Reviewers noticed it was, at best, not ready.
Devin turns out to have not been entirely forthright with its demos.
OpenAI fired two employees who had been on its superalignment team, Leopold Aschenbrenner and Pave... | https://www.lesswrong.com/posts/FAnxq8wFpfqGjeetC/ai-60-oh-the-humanity |
# Cooperation is optimal, with weaker agents too - tldr
*It's a ‘*[*superrational*](https://www.lesswrong.com/tag/superrationality)*’ extension of the* [*proven optimality of cooperation in game theory*](https://www.youtube.com/watch?v=mScpHTIi-kM)
*\+ Taking into account asymmetries of power*
***// Still** AI r... | https://www.lesswrong.com/posts/Sh4Js5koiaZCP2RsG/cooperation-is-optimal-with-weaker-agents-too-tldr |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.